This project was designed to put previos ML knowledge into practice. The game itself was developed with Pygame, and the code has an Interface (Abstract Class) implemented that allows the game to be playable or the agent to be trainable. The technology used for the AI is a Reinforcement Learning model, specifically name Deep Q Learning. The architecture consist in an input layer of 11 nodes conected to a single hidden layer of 1024 nodes and finally the conected to the output layer of 3 nodes. The model was done with Pytorch.
This project is intended for educational purposes only.